Repository navigation

Feature/207 kllama gguf - #307

Merged

michalharakal merged 22 commits into

developfrom

feature/207-kllama-gguf

Jan 18, 2026

Merged

Feature/207 kllama gguf#307
michalharakal merged 22 commits into
developfrom
feature/207-kllama-gguf

Commits on Jan 18, 2026
- Add initila version of port KMP llama to SKaiNET. Implements: #207
  michalharakal
  committed
  7269c2d
- Include module to project
  michalharakal
  committed
  654ee97
- Fix compile error
  michalharakal
  committed
  ba48f39
- Add initial version of port of llama2 from llama2-kmp project
  michalharakal
  committed
  7453184
- Add initial version simple wasm/js sample
  michalharakal
  committed
  fc0756e
- Add tokenizer implementaion
  michalharakal
  committed
  ca314c6
- kllama run s antive on mac with karphaty's model
  michalharakal
  committed
  0908590
- kllama workin in macosx cli
  michalharakal
  committed
  82f2d22
- Improve kllame loading of gguf files. Fix bugs in quantization and mapping
  michalharakal
  committed
  b330407
- Add GGUF embedded tokenizer support
  
  michalharakal
  and
  claude
  committed
  0d5901f
- Add shadow JAR support for JVM fat JAR build
  
  michalharakal
  and
  claude
  committed
  96e09e2
- Remove Karpathy .bin format support, simplify to GGUF-only
  
  michalharakal
  and
  claude
  committed
  fb7716c
- Add KLlama enterprise roadmap document
  
  michalharakal
  and
  claude
  committed
  4d21e62
- Add multi-backend storage abstraction architecture to roadmap
  
  michalharakal
  and
  claude
  committed
  69d3052
- Add mmap support for zero-copy GGUF tensor loading
  
  michalharakal
  and
  claude
  committed
  d3427ce
- Update roadmap: mark mmap implementation as completed
  
  michalharakal
  and
  claude
  committed
  32381ad
- Add BitNet/TQ1_0 support to roadmap as high priority
  
  michalharakal
  and
  claude
  committed
  c672829
- Add BitNet/Ternary quantization support (TQ1_0/TQ2_0)
  
  michalharakal
  and
  claude
  committed
  d74517e
- Add Q8_0/Q4_K quantized inference and off-heap KV cache
  
  michalharakal
  and
  claude
  committed
  7509509
- Add integration tests for quantized inference and KvCache
  
  michalharakal
  and
  claude
  committed
  7408086
- Update roadmap: mark Phase 1 as completed
  
  michalharakal
  and
  claude
  committed
  180a2cb
- Add Phase 1 verification tasks to roadmap
  
  michalharakal
  and
  claude
  committed
  c02515c

Commits

Commits on Jan 18, 2026